移动粒子半隐式方法的计算效率优化

针对无网格法计算效率普遍不高的局限性,以移动粒子半隐式(MPS)方法为例,将计算耗时较多的程序模块作为优化对象,以降低计算等级和减少计算次数为优化手段,从搜索算法、数学运算和计算实现形式3个主要方面进行了改进.提出了用低级运算来代替高级运算、用方位比较来代替距离比较的位置比较消冗法来优化搜索算法,消除了大部分无效计算,大大降低了计算的复杂程度和计算量,使得该方法的计算效率得到显著提高,增强了MPS方法的应用范围和工程时效性.该算法具有非常好的普适性,容易在类似算法中移植,还有随着计算规模的增大优化效果更加明显的特点.同时,发现多种优化方式的组合使用并非效果叠加,需考虑它们的兼容关系.

Optimization Study on Computational Efficiency of Moving Particle Semi-Implicit Method

The moving particle semi-implicit(MPS) method was improved to raise the computational efficiency of meshless methods from the search algorithm,the mathematical manipulation and the calculation execution.In the improvement,the program modules which consume the most computing time in the calculation were regarded as the optimization objects,and reducing the calculation level and the calculation amount was viewed as the optimization means.A new search algorithm was proposed to eliminate most of the useless calculation and reduce the complexity and the calculation amount by adopting low level mathematical manipulations and the position comparison.The results show that the MPS method improved has higher computational efficiency and wider application in engineering.It is suggested that the compatibility between different optimization methods should be considered because their combination may not yield the superposition of effects.